Use family-record discrepancies to refine an ancestor search
At a glance
A genealogist used ChatGPT to propose Scottish research leads, then corrected its record-reading mistakes against original evidence. This workflow is listed under Family History, Research, Document Verification.
- Agents / tools
- ChatGPT, Genealogical records
SourceBuilding a tool for these workflows? See weekly rates.
What they did
1) Gather the ancestor's known marriage and burial details, occupation, children's names, and the disputed parentage. 2) Give ChatGPT those facts and ask it to investigate the parents. Compare proposed identities with the known dates instead of accepting matching names. 3) Follow its occupational-record lead in The Dictionary of Edinburgh Wrights and Furniture Makers and inspect the entries yourself. 4) Add the marriage parish, Lady Yester's, and the burial transcript's address. Use the resulting parish, date, and surname-variant suggestions to narrow further searches. 5) When supplying a register screenshot, check each transcribed name and date against its actual row, correct mismatches, and treat the remaining families as research candidates.
How it went
The author found a relevant 1825–1827 occupational reference, but did not establish the parents.
Worth knowing
ChatGPT shifted induction dates between rows in a burgess list. Paywalled records still required the researcher's own searches.
